CN109657070B - A Construction Method of Terminal Assisted SWOT Index System - Google Patents

Abstract

The invention relates to the field of terminal centralized storage, in particular to a terminal auxiliary deviceSWOTThe construction method of the index system comprises the following steps: step (a)S100, extracting keywords of a text data set based on a terminal; step (a)S200 keyword clustering andSWOTmapping an index system; step (a)S300, generating an index system weight suggestion. The automatic extraction and clustering of the keywords effectively saves the human resources of the expert and avoids to a certain extentSWOTInfluence of human interference factors in the system construction process.

Description

A Construction Method of Terminal Assisted SWOT Index System

technical field

The invention relates to the field of terminal centralized storage, in particular to a method for constructing a terminal-assisted SWOT index system.

Background technique

SWOT analysis method (Strengths: internal advantage factors, Weakness: internal weakness factors, Opportunities: external opportunity factors, Threats: external threat factors) is a classic competitive intelligence analysis tool, developed by K.J. Andrews of Harvard Business School in 1971 In his book "Corporate Strategy Concepts". The main content of this method is to carry out extensive investigation and information collection around the analysis target, and then analyze the collected information to judge the external opportunities and threats affecting the target, and the internal advantages and disadvantages of target implementation. The SWOT analysis method can not only conduct simple preliminary analysis, qualitatively understand the overall situation of the analysis target, but also can realize the strategic strategy formation, implementation or control decision of the target.

Since the SWOT analysis method starts from the overall analysis of the target, it can clearly list the advantages, disadvantages, opportunities and threats that affect the implementation of the target, and make a comprehensive analysis to clarify the complex factors that affect the implementation of the target, so that decision makers can clearly grasp the target Risks and opportunities that may exist in the implementation, so as to improve the accuracy of decision-making. Therefore, the SWOT analysis method has become the most commonly used analysis tool in the management and decision-making of modern government departments and enterprises, and has been widely used and researched.

Based on the above technical problems, it is necessary to design a new construction method of terminal-assisted SWOT index system.

Contents of the invention

The purpose of the present invention is to provide a method for constructing a terminal-assisted SWOT index system.

In order to solve the above-mentioned technical problems, the present invention provides a method for constructing a terminal-assisted SWOT indicator system, including:

Step S100, extracting keywords from the text data set;

Step S200, keyword clustering and SWOT indicator system mapping; and

Step S300, generating weight suggestions for the index system.

Further, the method for extracting keywords from the text data set in the step S100 includes:

Step S101, filter stop words, after performing Chinese word segmentation on the collected text data set, filter the stop words in the text data by accumulating and selecting the stop word list formed;

Step S102, filtering specific words, searching for words through a search engine, and judging that the words whose search results are less than a threshold are specific words, and then filtering the specific words;

Step S103, keyword extraction, keyword extraction is performed through the improved TF/IDF algorithm.

Further, the improved TF/IDF algorithm is:

Formula 1

Formula 2 W _i ={W|TF/IDF(w _i )>η};

Formula 3 W=∪W _i ;

Formula 4

In the formula, TF/IDF( _wi ) is the TF/IDF weight of word w in the text data labeled i; TF( _wi ) is the frequency of word w appearing in the text data labeled i; N is the text The number of text data contained in the data set; d is the number of text data containing word w;

The method for keyword extraction by the improved TF/IDF algorithm comprises:

Calculate the TF/IDF weight of the keywords contained in each text data in the text data set by formula 1;

Sort by size according to the TF/IDF weight of keywords in each text data;

The keyword of extracting weight is greater than threshold value n forms the keyword collection Wi of the text data that label is i, and the Wi collection of all text data is summarized as the keyword W collection of text data set;

Calculate the ratio C for pairwise pairing of keywords in the W set;

In Equation 4, TF _sum (W _a ) refers to the cumulative sum of the frequency of a certain keyword a appearing in the W set, TF _sum (W _b ) refers to the cumulative sum of the frequency of a certain keyword b appearing in the W set, G(W _a ) refers to the number of search page results obtained by the keyword a in the search engine; G(W _b ) refers to the number of search page results obtained by the keyword b in the search engine; the ratio C is a pair of keywords a and b The ratio of the product of the TF _sum value and the G value, and the keywords in the W set are sorted according to the result of the ratio, and displayed in order to correct the keywords.

Further, the method for keyword clustering and SWOT index system mapping in the step S200 includes:

Step S201, according to the Chinese classification thesaurus, realize the initial classification of keywords, compare the Chinese classification thesaurus, classify the keywords extracted from the current text data set, and establish the initial keyword classification structure;

Step S202, for the remaining keywords that cannot be classified in the Chinese classification subject vocabulary after the preliminary classification, use the K_MEANS clustering method to cluster the remaining keywords according to the degree of similarity of words as the distance measure between words and words ;

Step S203, after the terminal-assisted clustering is completed, the clustered keywords are classified and displayed and corrected;

Step S204, after repeated iterations of keyword clustering and correction of the clustered keyword classification, according to the classification information of the clustered word categories, the word categories are mapped to corresponding indicators, namely

Establish an indicator system for SWOT analysis.

Further, the method for generating the weight suggestion of the index system in the step S300 includes: selecting factors that affect the judgment of the weight of the index system;

The factors affecting the weight judgment of the indicator system include:

The word volume of the keywords contained in the part of speech: by analyzing the number of keywords contained in each part of speech generated during the keyword clustering process, to determine the weight of the index generated by the map of the part of speech, that is, the corresponding index of the part of speech with more keywords The greater the weight;

The word frequency of the keywords contained in the part of speech: the cumulative sum of the frequencies of all keywords contained in the part of speech appearing in the text data set; and

The timeliness of keywords included in a part of speech: The word frequency statistics in the time dimension of keywords included in a part of speech show the degree of attention to the keyword in the time dimension.

Further, the method for generating the weight suggestion of the index system in the step S300 also includes: constructing a formula for generating the weight suggestion of the index system based on the factors that affect the weight judgment of the index system, namely

In the formula, R(W) is the index weight suggestion corresponding to a part of speech; i from 1 to k is the number of keywords contained in the part of speech, and all keywords in the class are calculated in turn; j from 1 to d is the number of keywords included in the part of speech For the text data of a certain word w in the part of speech, calculate all the text data containing the word w in turn; traverse the text data containing the word w, and calculate the time decay function of the jth word w containing the word respectively; TF(w _j ) is The frequency of word w appearing in text data j; e ^-μ(t-tc) is the time decay function; μ is the decay constant; t is the time when the text data appears; tc is the current moment;

After calculating the R(W) weight suggestion value of each part of speech, an index weight suggestion is generated.

The beneficial effect of the present invention is that the present invention is based on the keyword extraction of the text data set by the terminal, and maps the keyword clustering to the SWOT index system; and generates index system weight suggestions, thereby realizing automatic keyword extraction and clustering.

Description of drawings

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

Fig. 1 is a flow chart of the construction method of the terminal-assisted SWOT index system involved in the present invention.

Detailed ways

The present invention is described in further detail now in conjunction with accompanying drawing. These drawings are all simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, so they only show the configurations related to the present invention.

Example 1

Fig. 1 is a flow chart of the construction method of the terminal-assisted SWOT index system involved in the present invention.

As shown in Figure 1, this embodiment provides a method for constructing a terminal-assisted SWOT indicator system, including:

Step S100, based on the keyword extraction of the text data set by the terminal;

Step S200, keyword clustering and SWOT indicator system mapping; and

Step S300, generating weight suggestions for the index system;

In this implementation, the terminal can, but is not limited to, use a computer to assist in the construction of the SWOT index system; the automatic extraction and clustering of keywords effectively saves human resources for experts and avoids the process of building the SWOT system to a certain extent Influenced by human interference factors.

In this embodiment, the step S100, the terminal-based method for extracting keywords from the text data set includes: step S101, stop word filtering, after performing Chinese word segmentation on the text data set collected by the terminal, the words formed by accumulating and selecting Stop word list, filter stop words in the text data, the stop words are generally modal auxiliary words, function words and quantifiers, etc.; step S102, filter specific words, search words by search engine, and the search results are less than the threshold The word is judged to be a specific word, and then the specific word is filtered, and the specific word is generally a specific word with strong directivity such as place name and person name; unlike stop words, it is difficult to filter specific words through a customized vocabulary, In related research work, many practical word classification reasoning is used to classify specific words and judge whether words are place names, personal names, etc., but this kind of inference has certain unreliability; using search engines, such as Google, Baidu, etc., To judge specific words; for example, Google will display the number of search results every time you search, and you will get fewer search pages if you use specific words to search. Therefore, for words with less than a certain threshold in search results, you can judge them as specific words. be filtered; Google's retrieval of specific words can be automatically completed by Google's algorithm; step S103, keyword extraction, keyword extraction is carried out by the improved TF/IDF algorithm; TF/IDF algorithm is the current mainstream keyword extraction algorithm, TF (Term Frequency: word frequency), refers to the number of times a word appears in a text, IDF (Inverse Document Frequency: inverse document frequency).

In this embodiment, what needs to be implemented is to extract the keywords in the set for the entire text data set. The traditional TF/IDF algorithm is to extract the keywords in the document for a certain document, so for the traditional TF/IDF The IDF algorithm is improved; the improved TF/IDF algorithm is:

Formula 1

Formula 2 W _i ={W|TF/IDF(w _i )>η};

Formula 3 W=∪W _i ;

Formula 4

In the formula, TF/IDF( _wi ) is the TF/IDF weight of word w in the text data labeled i; TF( _wi ) is the frequency of word w appearing in the text data labeled i; N is the text The number of text data contained in the data set; d is the number of text data containing word w;

The method for extracting keywords through the improved TF/IDF algorithm includes: calculating the TF/IDF weight of keywords contained in each text data in the text data set by formula 1; according to the TF of keywords in each text data The /IDF weights are sorted by size; the keywords whose weight is greater than the threshold η are extracted to form the keyword set Wi of the text data labeled i, and the Wi collections of all text data are summarized as the keyword W collection of the text data set; for W The keywords in the set are paired in pairs, and the ratio C is calculated; in formula 4, TF _sum (W _a ) refers to the cumulative sum of the frequency of a certain keyword a in the W set, and TF _sum (W _b ) refers to the frequency of a certain keyword b in the W set. The cumulative sum of frequencies that appear in the W set, G(W _a ) refers to the number of search page results obtained by the keyword a in the search engine; G(W _b ) refers to the search page results obtained by the keyword b in the search engine The number of results; the ratio C is the TF _sum value and G value of a pair of keywords a and b (the G value is the same as the expression form of TF _sum , referring to the number of search page results obtained by a pair of keywords in the search engine) The ratio of the product, and sort the keywords in the W collection according to the result of the ratio, and display in order to correct the keywords.

In this embodiment, the method of keyword clustering and SWOT index system mapping in step S200 includes: step S201, according to the Chinese classification thesaurus, realize the initial classification of keywords, compare the Chinese classification thesaurus, and Classify the keywords extracted from the current text data set, and establish an initial keyword classification structure; step S202, for the remaining keywords that cannot be classified in the Chinese classification subject list after the preliminary classification, according to the degree of similarity of the words As the distance measure between words, the K MEANS clustering method is used to cluster the remaining keywords; step S203, after the terminal-assisted clustering is completed, the clustered keywords are classified and displayed and corrected, and the corrected The method can be, but not limited to, manually corrected; step S204, after repeated iterations of keyword clustering and correction of the clustered keyword classification, according to the classification information of the clustered word categories, the word categories are mapped to corresponding indicators , that is, to establish the index system of SWOT analysis.

In this embodiment, the step S300, the method for generating the weight suggestion of the index system includes: selecting factors that affect the weight judgment of the index system; the support degree of each index in the index system for the analysis result is different, that is, some indexes are the main factors , while some indicators are secondary factors. In this embodiment, weight suggestions are generated through three factors that affect the weight judgment of the index system; the factors that affect the weight judgment of the index system include:

The word volume of the keywords contained in the part of speech: by analyzing the number of keywords contained in each part of speech generated during the keyword clustering process, the weight of the index generated by the mapping of the part of speech is judged, that is, the index corresponding to the part of speech with more keywords The greater the weight;

Word frequency of keywords contained in a part of speech: In addition to the number of keywords, the frequency of keywords contained in a part of speech is also the basis for judging the weight of the index mapped to the part of speech. Cumulative sum of frequencies appearing in the data set;

The timeliness of the keywords contained in the word category: the frequency of keywords appearing in a certain period of time, the text data sets collected through open source data all have time attributes, and the words in the text data are also attached with the time attributes of the text data , the time attribute of words is not considered when analyzing and extracting keywords, but the word frequency statistics of the keywords contained in a part of speech in the time dimension show the degree of attention of the keyword in the time dimension, that is, the words contained in the part of speech The timeliness of keywords is also an element in judging the weight of their corresponding indicators.

In this embodiment, the step S300, the method for generating the weight suggestion of the index system further includes: constructing a formula for generating the weight suggestion of the index system based on the factors that affect the weight judgment of the index system, namely

In the formula, R(W) is the index weight suggestion corresponding to a part of speech; i from 1 to k is the number of keywords contained in the part of speech, and all keywords in the class are calculated in turn; j from 1 to d is the number of keywords included in the part of speech For the text data of a certain word w in the part of speech, calculate all the text data containing the word w in turn; traverse the text data containing the word w, and calculate the time decay function of the jth word w containing the word respectively; TF(w _j ) is The frequency of word w appearing in text data j; e ^-μ(t-tc) is the time decay function; μ is the decay constant; t is the time when the text data appears; tc is the current moment; calculate the R(W ) to generate an indicator weight suggestion after the weight suggestion value.

In summary, the present invention extracts keywords from text data sets, maps keywords clustering and SWOT index system, and finally generates index system weight suggestions, realizes automatic keyword extraction and clustering, and effectively saves experts human resources, and to a certain extent avoid the influence of human interference factors in the SWOT system construction process; and the present invention can also effectively save the workload of intelligence analysts, and can reduce the SWOT index system construction process Interfering factors, which also have a facilitative significance for the application of SWOT analysis.

The first iteration is used to obtain the keywords related to the SWOT analysis target and cluster the keywords. The goal of the second iteration is to map the keyword classes into SWOT evaluation indicators. The last part, after the SWOT evaluation indicators are generated, is passed Algorithms generate recommendations for SWOT indicator weights.

Inspired by the above-mentioned ideal embodiment according to the present invention, through the above-mentioned description content, relevant workers can make various changes and modifications within the scope of not departing from the technical idea of the present invention. The technical scope of the present invention is not limited to the content in the specification, but must be determined according to the scope of the claims.

Claims (3)

1. The construction method of the terminal-assisted SWOT index system is characterized by comprising the following steps of:

step S100, extracting keywords of a text data set;

step S200, keyword clustering and SWOT index system mapping; and

step S300, generating an index system weight suggestion;

the method for extracting the keywords of the text data set in the step S100 comprises the following steps:

step S101, filtering stop words, namely after Chinese word segmentation is carried out on an acquired text data set, filtering the stop words in the text data by accumulating stop word lists formed by selection;

step S102, filtering specific words, searching the words through a search engine, judging the words with search results less than a threshold value as specific words, and filtering the specific words;

step S103, extracting keywords, namely extracting the keywords through an improved TF/IDF algorithm;

the improved TF/IDF algorithm is:

1 (1)

2W _n ＝{W|TF/IDF(w _n )＞η}；

3W = u W _n ；

4. The method is to

Wherein TF/IDF (w _n ) TF/IDF weight for word w in text data numbered n; TF (w) _n ) Frequency of occurrence of the word w in the text data with the number n; n is the number of text data contained in the text data set; d is the number of text data containing word w;

the method for extracting the keywords through the improved TF/IDF algorithm comprises the following steps:

calculating TF/IDF weight of keywords contained in each text data in the text data set through the method 1;

sorting according to the size of the TF/IDF weight of the keywords in each text data;

extracting keywords with weights larger than a threshold value eta to form a keyword set W of text data with a label of n _n W of all text data _n The collection summary is a keyword W collection of the text data collection;

aiming at pairwise pairing of keywords in the W set, calculating a ratio C;

TF in 4 _sum (W _a ) Refers to the cumulative sum of the frequency of occurrence of a certain keyword a in a W set, TF _sum (W _b ) Refers to the cumulative sum of the frequency of occurrence of a certain keyword b in a W set, G (W) _a ) The search page result number obtained by the keyword a in the search engine is referred to; g (W) _b ) The result number of the search page obtained by the keyword b in the search engine is referred to; the ratio C is TF of a pair of keywords a and b _sum The ratio of the product of the value and the G value, and the keywords in the W set are ordered according to the result of the ratio and displayed in sequence to correct the keywords;

the step S200, the method for mapping the keyword clusters and the SWOT index system comprises the following steps:

step S201, according to the Chinese classification subject word list, the primary classification of the keywords is realized, the keywords extracted from the current text data set are classified by contrasting with the Chinese classification subject word list, and an initial keyword classification structure is established;

step S202, clustering the residual keywords by adopting a K_MEANS clustering method according to the word proximity degree serving as a word-word distance measure for the keywords which cannot be classified correspondingly in the Chinese classification subject word list after preliminary classification;

step S203, after the terminal auxiliary clustering is completed, the clustered keywords are displayed and corrected in a classified manner;

step S204, after repeated iteration of keyword clustering and classification correction of clustered keywords, mapping the word classes into corresponding indexes according to classification information of the clustered word classes, namely

And establishing an index system of SWOT analysis.

2. The construction method according to claim 1, wherein,

the method for generating the index system weight suggestion in the step S300 includes: selecting factors influencing weight judgment of an index system;

the factors influencing the weight judgment of the index system comprise:

word quantity of keywords contained in word class: judging the index weight generated by mapping each word class by analyzing the number of keywords contained in each word class generated in the keyword clustering process, namely, the index weight corresponding to the word class with the larger number of keywords is larger;

word frequency of keywords contained in word class: accumulating and summing the frequencies of all keywords contained in the word class in the text data set;

the timeliness of keywords contained in the word class: word frequency statistics in the time dimension for keywords contained in a word class show how much attention is paid to the keywords in the time dimension.

3. The construction method according to claim 2, wherein,

the method for generating the index system weight suggestion in the step S300 further includes: constructing a generation formula of index system weight suggestion based on factors influencing index system weight judgment, namely

Wherein R (W) is an index weight suggestion corresponding to a word class; i is the number of keywords contained in the word class from 1 to k, and all keywords in the word class are calculated in sequence; j is text data comprising a word w in the word class from 1 to d, and all text data comprising the word w are sequentially calculated; traversing text data containing word w, and respectively calculating the j-th time decay function containing word w; TF (w) _j ) Is the frequency with which word w appears in text data j; e, e ^-μ(t-tc) As a function of time decay; mu is the decay constant; t is the time when the text data appears; tc is the current time;

an index weight suggestion is generated after calculating an R (W) weight suggestion value for each word class.

Images